- אירוע כבר עבר.
Optimizing Privacy Choices
יוני 4, 2019 @ 12:00 pm - 1:00 pm
Speaker: Dr. Ron Hirschsprung, Department of Industrial Engineering, Tel Aviv University
Time: 12:00 – 13:00
Place: Building #3A, upper conference room, floor 3 (floor 2 in the elevator), Ariel University, Ariel
Information sharing services such as online social networks, eCommerce, and cloud applications are at the heart of contemporary information systems. Inherently they introduce a tradeoff between benefits and privacy loss. Many systems provide some mechanism to control this tradeoff, however, the way sharing controls are designed (choice architectures), have a strong effect on the decisions of users. For example, it has shown that Facebook users fail to correctly manage their privacy settings. Counter-intuitively, increasing the number of choices leads to weaker preferences and the task of configuring privacy can involve significant cognitive load for the user.
Thus, we look for a methodology and a theoretical foundation to reduce, optimize and evaluate the configuration spaces for privacy choices. We conducted three main researches: a) Optimizing choices based on users’ preferences. We developed a methodology of reducing the configuration space to a small set of canonical options that satisfy a maximal portion of the users (coverage). We empirically evaluate our methodology (implemented as an algorithm named SCON-UP – Simplified CONfiguration of User Preferences) using data collected in two user studies (n=121 and n=352). The results show that for a reduced set of 3 canonical configurations above 80% of the population is covered. b) Evaluating choice architecture. We implemented SCON-UP to the domain of Facebook, by using a real data of n=266 users who published 21,950 posts. Our results indicate that on average, while Facebook provides coverage of round 76%, the optimized set reaches 85%. c) Optimizing choice architectures based on the value of privacy. We developed a methodology named VOPE (Value Of Privacy Estimator), that can accommodate also average utilities and social fairness in the objective function. The methodology is based on estimating an intrinsic value (e.g., Dollars) of privacy as perceived by the user. To evaluate VOPE, we conducted a user study with n=195 participants. VOPE is based on a converging process, and on average the value converged (i.e., provided) on 68.5% of the times. We validate VOPE results by comparing them to another study done by other researcher and by implementing a validation methodology relying on another independent online study.
We demonstrated how our SCON-UP and VOPE methodologies can be applied for evaluating and optimizing existing choice architectures in information systems, effectively by reducing the configuration space. Furthermore, by using VOPE approach the objective function of the choice architecture optimization process can be multi-parametric. Our results establish a theory of optimizing the trade-off of social welfare, coverage, and fairness in choice architecture that involved privacy. Therefore, enables a designer or a regulator to balance the trade-off between usability, utility, and fairness in choice architecture. This work can bridge between policies creators, aimed to protect users’ privacy, and choice architecture designers usually aimed to increase usability. The methods can be extended and applied to other fields in which there is an extensive space of choices that represent a trade-off between different costs and benefits to optimize and reduce the configuration space.
Ron Hirschprung has rich and extensive experience in the industry of Information Systems, as a founder, director, and manager of various companies. His main research interest is on the development of methodologies to bridge the fuzzy world of human conceptions with the deterministic world of computer systems, while focusing on privacy issues. Ron obtained his Ph.D. degree in Industrial Engineering from Tel Aviv University.